ABSTRACT
The Prince Edward Island Department of Transportation, Infrastructure and Energy (TIE) is responsible for the densest network of roadway and bridges per capita in Canada. Over the last 12 years, TIE has adopted bridge management processes including implementing an advanced bridge management system (BMS). This has enabled the department to make significant improvements in the management of inventory and inspection data, the inspection process, tracking and reporting condition through a standard condition index, performing asset valuation, and in handling of inspection photos, videos, plans and other documents.The TIE bridge manager currently reports to senior management on the overall bridge condition index (BCI) and risk assessment through the bridge criticality and urgency (BCU) index on a network level. The report includes historical tracking of these indices, a 5 year capital plan and a forecast of condition which is based on the BMS’s powerful analytical tools. These forecasts rely on the deterioration models that are used by the BMS so it was of interest to the department to study the effectiveness of the deterioration models in predicting real-world deterioration.TIE have (12) years of and inspection cycles and condition forecasting from the BMS, which is reported on an annual basis to senior management. The forecasts that are selected cover three scenarios: 1) Do Nothing; 2) Unlimited Budget and 3) a user defined Constrained Budget. The BMS uses default Markovian deterioration model to predict deterioration of all bridge and culvert elements/materials. Recently, with this many years of condition data available it became possible to investigate the field inspection derived deterioration as recorded by the condition state inspection and compare this to the default (expert knowledge) deterioration models. The BMS includes a Bayesian Updating procedure that can be used to research the field inspection derived Markov deterioration rates, compare to the expert models, and then develop new models.This paper will present the results of the deterioration model Bayesian updates and verification.
